Making virtual brain models more biologically realistic

A new mean-field model connects ion exchange at the cellular level (left) with the collective activity of neuronal populations (centre) and large-scale brain networks (right). IMAGE: Rabuffo G. et al., eLife, 2026)

Mean-field models are a powerful tool for understanding collective brain activity. A new model extends this approach by incorporating the exchange of ions between neurons and their surroundings, bringing important biological mechanisms into simulations of neuronal populations.

How can researchers simulate the activity of the whole brain without modelling every neuron in full detail? A new study, supported in part by EBRAINS 2.0 and published in eLife, explores one part of the answer: retaining key biological mechanisms while simplifying the mathematics needed to model large neuronal populations.

The brain's electrical activity depends on the movement of charged particles, called ions, across cell membranes and through the fluid surrounding neurons. These processes also influence one another: changes in ion concentration can affect how neurons fire and, in some circumstances, contribute to abnormal brain activity, including epileptic seizures.

Capturing these interactions in large-scale computer simulations is challenging. Simulating many individual neurons in detail can be computationally demanding. Researchers therefore often use “mean-field models”, which approximate the collective behaviour of a neuronal population using a smaller set of mathematical equations.

In this study, researchers from France, Spain and Finland present a new mean-field model derived from a network of Hodgkin–Huxley-type point neurons. These are mathematical models that describe how electrical signals arise from the movement of ions across neuronal membranes. The new model uses a simplified description of neuronal populations while retaining equations for ion exchange between the inside of cells and the surrounding extracellular space.

Connecting cellular mechanisms with network activity

“The team shows that the model can reproduce several patterns of activity also seen in simulations of larger neuronal networks and in in vitro recordings from mouse brain slices,” says Spase Petkoski, co-senior author of the study and researcher at the Institut de Neurosciences des Systèmes, Aix-Marseille University. “In both cases the model captures how gradual changes in potassium concentrations can influence, either healthy or pathological, faster electrical activity.”

The approach builds on the widely used mean-field and neural mass modelling methods, and its contribution is to incorporate ion-exchange mechanisms into a reduced model derived from Hodgkin-Huxley-type neurons.

This offers researchers another way to investigate how changes in the brain’s chemical environment influence collective neuronal activity. The model could potentially support future investigations into how changes in ion concentrations affect network dynamics, including processes relevant to epilepsy.

The work is relevant to EBRAINS' efforts to advance multiscale brain modelling, connecting biological processes at different levels of brain organisation through computational tools. The study's co-senior author, Viktor Jirsa, is Chief Science Officer of EBRAINS. The research received funding through the EBRAINS 2.0 project and the Virtual Brain Twin project. The model's publicly available code also includes an implementation for The Virtual Brain, a platform for brain-network modelling.

Author: Helen Mendes

Contact: press@ebrains.eu